Audio Ad Production Cost: Script, Voice, Music, and Revisions Explained

What you are actually buying

A finished MP3 or WAV is only the last artifact. The work and rights behind it determine the price.

This is why an “all-in” quote is not necessarily complete. It is complete only when the scope underneath it is legible.

Keep production cost separate from the media buy

Media spend buys delivery: impressions, reach, frequency, clicks, or another campaign outcome. Production spend creates and licenses the advertisement that will be delivered. A platform may subsidize production because it expects the advertiser to buy media there. A studio quote, by contrast, may price a standalone master intended for several publishers.

The practical rule is to maintain two budget columns:

1. Creative production: everything required to make, license, revise, and deliver the ad.
2. Media: platform budget, buying fees, audience data, trafficking, measurement, and optimization.

When a platform offers inexpensive or included creative, confirm rather than assume whether you can download the final master, reuse it elsewhere, retain the dry voice track, change the music, create cutdowns, or keep using the talent after the campaign ends. Low in-platform production cost can be perfectly rational for a narrow test. It is not automatically comparable with a quote for reusable campaign assets.

Script cost: you are paying for decisions, not word count

A 30-second script is short, but short copy can require substantial decision-making. There are at least four different script scopes:

• Client-supplied script: the producer records exactly what you approve.
• Script polish: the producer tightens timing, pronunciation, CTA, and spoken rhythm.
• Brief-to-script: the writer turns an offer, audience, and proof points into one finished route.
• Concept development: the team proposes several creative approaches before writing one or more scripts.

Those should not have the same price. “Writing included” is too vague unless the quote gives the number of routes, drafts, stakeholder rounds, and final scripts.

Timing also changes the work. Spotify’s current audio specifications recommend no more than 65 words for a 30-second ad. The same page specifies a 30-second maximum for standard music and streaming-inserted podcast ads. (Spotify Advertising, “Audio ad specs & requirements”) A 49-word script can leave room for a useful pause, a natural brand read, and a clear URL. A 75-word script may technically fit with a fast performer but can remove the very timing choices that make the message understandable.

Approve these before recording:

• the exact offer, date, price, disclaimer, CTA, and destination;
• phonetic pronunciation for the brand, product, people, and places;
• which words need emphasis and where silence is intentional;
• the legal or compliance owner for regulated or comparative claims;
• the final word count and target duration when read naturally.

A late copy change is not “just editing” if it requires the actor to return to the microphone. The cheapest revision is a firm script approval gate before casting and recording.

Voice cost: separate the performance from its use

A professional voice quote often contains two economic ideas even when they appear as one total:

1. Session or performance fee: compensation for preparing, recording, directing, and delivering the performance.
2. Usage fee: permission to use that performance in defined advertising.

Voices describes paid promotion—including TV, radio, and online ads—as broadcast work and says pricing is driven by usage factors such as location, duration, and platform. Non-broadcast narration is commonly priced differently. (Voices, “Voice Over Rate Guide”) That distinction matters because a 30-second internal training narration and a 30-second national paid ad are not equivalent uses of the performer’s voice.

A voice booking should spell out:

Union and non-union quotes are not interchangeable

SAG-AFTRA states that its voiceover performers work under several collective bargaining agreements that provide guaranteed minimum rates, health and retirement contributions, residuals, and other compensation. It also directs producers to the contract area appropriate to the project. (SAG-AFTRA, “Voiceover”)

That does not mean every union job has one universal price. It means a buyer comparing union and non-union bids should account for the applicable agreement, signatory process, contributions, use payments, and administrative obligations rather than comparing only the actor’s apparent session fee. Ask the producer to quote the current applicable contract instead of copying a stale rate from a search result.

Human voice and AI voice require explicit rights language

An AI voice may reduce casting and rerecording time, but “synthetic” does not remove the need to understand consent, provenance, and permitted use. The National Association of Voice Actors’ current AI rider prohibits using a talent’s recording beyond the initial agreement for synthetic or cloned voices, machine learning, or AI training, and addresses transfer of original files without the talent’s knowledge and consent. (NAVA, “Synthetic Voice/AI Rider, version 4.5”)

For a human performance, put the no-training and no-cloning terms in writing. For a generated voice, document which provider supplied it, whether commercial paid advertising is allowed, whether the voice imitates a real person, how long the output may be used, and whether the provider can change or revoke the license. “AI voice included” is a production method, not a complete rights clause.

Music cost: “royalty-free” does not mean unrestricted

Music can be omitted, licensed from a library, commissioned as a custom bed, or developed into a reusable sonic identity. Those are different purchases.

The U.S. Copyright Office explains that a song contains two distinct copyrighted works: the underlying musical work and the sound recording. (U.S. Copyright Office, “How Songwriters, Composers, and Performers Get Paid”) When you use an existing recording, the safe commercial question is not merely “Do we have the song?” It is whether the license chain covers the composition and recording for the actual advertising use.

Stock libraries can simplify that chain, but their plans remain scope-specific. Soundstripe’s terms, for example, distinguish plan permissions for paid ads and broadcast use, limit some business distribution by market, and discuss public-performance licensing and cue sheets. (Soundstripe, “Terms and Conditions of Use,” updated November 24, 2025) PremiumBeat’s current single-track menu likewise lists different prices and permissions: $59 for its Standard commercial web/social tier, $199 for its Premium non-web advertisement tier, and $999 for expanded Business coverage. Those are one provider’s live prices, checked September 9, 2026—not a market average. (PremiumBeat, “Licensing”)

A useful music line item should answer:

• Is this a stock track, custom cue, licensed commercial recording, or platform library?
• Which media, territories, brands, and campaign versions are covered?
• Is the term perpetual for the finished project, or tied to a campaign period?
• Can the producer create cutdowns, loops, stems, and alternate mixes?
• Is paid advertising, radio, podcast, social, CTV, and client work included?
• Who keeps the license receipt and handles any content claim or cue sheet?

Custom music costs more because you are buying composition time, feedback, arrangement, production, and a negotiated rights package. A custom cue can also reduce future search and clearance work if the contract grants useful reuse rights. Do not call it a “buyout” unless the agreement defines exactly which copyrights or licenses transfer.

Editing, mixing, and mastering cost

Recording a clean read is not the same as delivering a traffickable ad. Post-production may include:

• selecting and assembling takes;
• removing clicks, noise, distracting breaths, and timing errors;
• fitting the read to a precise duration without making it sound rushed;
• editing music and sound effects around the message;
• equalization, compression, de-essing, level automation, and limiting;
• loudness measurement, file conversion, naming, and final quality control.

Platform specifications can create additional masters. Spotify currently accepts MP3, WAV, or OGG; specifies 44.1 kHz, stereo, a 192–320 kbps range, and a target of -16 LUFS within its stated tolerance. (Spotify Advertising, “Audio ad specs & requirements”) Another publisher may request a different format or loudness target. The producer should master to the delivery specification you provide, not to a generic “sounds loud” reference.

State the deliverables as files, not adjectives. For example:

• one 30-second stereo WAV master;
• one matching 320 kbps MP3 preview;
• one 15-second cutdown;
• one alternate-CTA master;
• dry voice, music-only mix, and separated stems;
• session files or an archived project, if required;
• a text file containing script, talent, music source, license, and usage dates.

Stems and source sessions are not automatically included. They can be valuable when you expect localization, offer changes, or new platform masters, so price them before the project closes.

Revision cost: define the kind of change

“Two revisions included” is incomplete because four very different events can be called a revision.

A round should mean one consolidated, time-stamped set of comments from the client’s authorized approver. It should not mean an unlimited sequence of messages from several stakeholders.

Also set a review window. A quote can include two rounds requested within ten business days, while changes requested six months later are a reopening. For voice pickups, state whether the actor must match the original session and how long that match window lasts. For music, distinguish a mix adjustment from choosing an entirely new track.

A revision contingency can be a useful budget line, but it should not hide undefined labor. The best quote tells you the included rounds, the hourly or fixed change-order price, and which events require renewed usage permission.

The same 30-second ad priced three ways

The following is an original fictional teaching example, not a vendor rate card or campaign result. All figures are in USD and exclude tax, media spend, union contributions, rush fees, translation, and legal review.

Brief: Northline Meal Kits wants one US English 30-second audio ad. The core offer and URL are approved. The brand may either run a short streaming test, build a one-year digital campaign, or prepare a broader North American audio package.

Approved 49-word script:

Weeknight dinner should not start with a scavenger hunt. Northline Meal Kits delivers pre-portioned ingredients and step-by-step recipes to your door. Pick meals for two or four, skip any week, and cook when it works for you. Get twenty dollars off your first box at NorthlineMeals.com. That is NorthlineMeals.com.

Three itemized quotes

What each total actually delivers

Scope A is rational when the offer is stable, the test is narrow, and the brand accepts rebuilding later. Scope B buys better direction, a longer usage term, reusable components, and one cutdown without funding a full audio identity. Scope C is not “a more expensive MP3.” It buys broader rights, more creative options, a custom cue, and a version system intended for several placements.

The correct choice depends on expected reuse. Paying $6,700 for one short test would be excessive. Paying $850 and then discovering that the voice, music, and source files cannot support a national campaign can also be expensive. Buy the smallest scope that covers the campaign you can reasonably foresee, then attach pre-priced renewal and expansion options.

Send every vendor the same filled quote request

Comparable quotes begin with a comparable brief. Here is the Northline request in a form a producer could price without guessing:

Project: Northline Meal Kits 30-second US English audio ad.
Script: Approved 49-word script supplied; quote script polish separately from concept development.
Voice: One human voice; warm, direct, conversational; quote session and usage separately.
Usage: Paid streaming audio in the United States for 13 weeks from first use; no exclusivity. Provide optional prices for 12 months and for adding podcast and terrestrial radio.
AI/data: No cloning, synthetic derivatives, training, voice conversion, or transfer into a dataset.
Music: Quote one cleared stock bed and an optional custom cue. State media, territory, term, and proof of license.
Deliverables: 30-second WAV to supplied platform specs, matching MP3, script/credits/license sheet; price 15-second cutdown, dry VO, stems, and alternate CTA separately.
Revisions: Identify included producer corrections, performance pickups, mix rounds, copy-change fees, and the review window.
Schedule: First review in five business days; final within two business days of consolidated notes.
Exclusions: Media spend, translation, legal review, taxes, rush work, and platform trafficking unless expressly added.

Ask each vendor to return the quote with a validity date, payment schedule, cancellation terms, talent hold or renewal terms, and unit prices for likely additions. That turns a mysterious total into a procurement decision.

Seven cost traps to catch before approval

1. The voice fee has no usage definition. “Commercial rights” is not a term, territory, platform list, or renewal price.
2. Music is described only as royalty-free. The plan may not cover paid ads, radio, client work, the territory, or the required duration.
3. The script is not frozen before recording. A stakeholder’s late offer change becomes a new session.
4. A revision round is undefined. Five reviewers can turn “one round” into weeks of fragmented notes.
5. Only one flattened master is delivered. A later CTA, language, or duration requires rebuilding because dry VO and stems were excluded.
6. A platform-created ad is treated as a portable asset without checking. Confirm downloads, reuse, music, voice, and modification rights in writing.
7. Media is buried in the production total—or vice versa. Keep the cost of making the ad separate from the cost of distributing it.

Price the scope before you price the file

The most useful audio ad quote is not the cheapest total. It is the one that makes the creative work, usage rights, music clearance, deliverables, and change rules visible.

Start with one approved brief. Separate session fees from usage. Name the music license. List every master and source asset. Define a revision round. Put media spend in its own budget. Then compare vendors line by line.

Once those fields match, the price difference becomes explainable—and you can choose whether you need a narrow test asset, a managed campaign package, or a reusable audio system instead of paying for surprises later.

Sources

• “Audio ad specs & requirements” — Spotify Advertising
• “Creative Services” — AudioGO
• “Audio Advertising Pricing” — AudioGO
• “Voice Over Rate Guide” — Voices
• “Voiceover” — SAG-AFTRA
• “Synthetic Voice/AI Rider, version 4.5” — National Association of Voice Actors, updated April 14, 2026
• “How Songwriters, Composers, and Performers Get Paid” — U.S. Copyright Office
• “Terms and Conditions of Use” — Soundstripe, updated November 24, 2025
• “Licensing” — PremiumBeat